June 20th, 2022


At this stage in the pandemic there is a view that the global spread of B.1.1.529 (Omicron), through its association with a relatively milder disease phenotype and, possibly, a potential to boost vaccine immunity, may herald the transition into a new, endemic relationship ( 28 ). The case for vaccine-mediated immune preconditioning as key mediator of the attenuated phenotype is complex: while functional neutralization by vaccine-primed sera is considerably blunted against B.1.1.529 (Omicron), three-dose vaccination efficacy against symptomatic disease holds up, in the 50-70% range ( 6 – 8 ).

It has thus been proposed that immune protection may be supported by maintenance of relatively high T cell response frequencies to viral epitopes unperturbed by loss of antibody epitopes ( 13 – 18 ). A rationale for this T cell mediated protection comes from animal studies showing the direct ability of SARS-CoV-2 specific T cells to curtail lung viral loads ( 29 ). This raised two key questions with respect to understanding and management of this wave: (i) considering the very diverse patterns of antiviral immunity shown by ourselves and others to be determined by differential immune imprinting, how would differences in antigen exposure through infection and vaccination alter immune responses against B.1.1.529 (Omicron) at the level of binding antibody and nAb, MBC and T cell responses? (ii) Is the immune response following infection during the B.1.1.529 (Omicron) wave primed and fully available to support protective immunity? We examined immunity to B.1.1.529 (Omicron) in a longitudinal HCW cohort, considering cross-reactive immunity primed by the varied spike exposures of 3-dose vaccination with or without hybrid immunity from any of the Wuhan Hu-1, B.1.1.7 (Alpha) or B.1.617.2 (Delta) infection waves, and then, the additive effective of actual infection during the B.1.1.529 (Omicron) wave. In the first part of this paper we report patterns of response in differentially imprinted, triple-vaccinated HCW. In the second part of the paper we consider immune responses in those who went on to suffer infection during the B.1.1.529 (Omicron) infection wave despite triple-vaccination. There were several unexpected findings. While it is known that cross-reactive antibody recognition is compromised by the mutations in B.1.1.529 (Omicron), it was surprising that this was so profoundly exacerbated by differential imprinting in those having had prior infection with either Wuhan Hu-1 or B.1.1.7 (Alpha). This adds an important dimension to global control of B.1.1.529 (Omicron) in light of the impact B.1.1.7 (Alpha) has had on the global pandemic: by May 2021 B.1.1.7 (Alpha) accounted for 67% of all cases across 149 countries ( 30 ). That previous SARS-CoV-2 infection history can imprint such a profound, negative impact on subsequent protective immunity is an unexpected consequence of COVID-19. While the notion that, generally, hybrid priming by infection and vaccination enhances immunity is widely agreed ( 22 ), imprinted patterns such as the specific combination of vaccination with infection during the first ancestral Wuhan Hu-1 wave followed by the B.1.1.529 (Omicron) wave requires an additional term—“hybrid-immune-damping.” Molecular characterization of the precise mechanism underpinning repertoire shaping from a combination of Wuhan Hu-1 or B.1.1.7 (Alpha) infection and triple-vaccination using ancestral Wuhan Hu-1 sequence, impacting immune responses to subsequent VOCs, will require detailed analysis of differential immune repertoires and their structural consequences. The impact of differential imprinting was seen just as profoundly in T cell recognition of B.1.1.529 (Omicron) S1, which was not recognized by T cells from any triple-vaccinated HCW who were initially infected during the Wuhan Hu-1 wave and then re-infected during the B.1.1.529 (Omicron) wave. Importantly, while B1.1.529 (Omicron) infection in triple-vaccinated previously uninfected individuals could indeed boost antibody, T cell and MBC responses against other VOC, responses to itself were reduced. This relatively poor immunogenicity against itself may help to explain why frequent B.1.1.529 (Omicron) reinfections with short time intervals between infections are proving a novel feature in this wave. It also concurs with observations that mRNA vaccination carrying the B.1.1.529 (Omicron) spike sequence (Omicron third-dose after ancestral sequence prime/boosting) offers no protective advantage ( 31 ). Initial studies using acute serum samples following B.1.1.529 (Omicron) infection had indicated poor immunogenicity and a tendency to elicit only Omicron-specific responses in the unvaccinated and broader responses in those imprinted following COVID-19 vaccination ( 32 , 33 ), including unexpected patterns of combinations that appeared to ablate neutralizing responses to previously seen VOC ( 33 ).

Our T cell analysis, which depended on processing of immunodominant epitopes from whole antigen, revealed a more profound deficit than others. Studies in which T cell responses of vaccinees against spike peptide megapools are screened show that, while there may be a 20% drop in response due to lost epitopes across the entire sequence, most of the response is maintained (13–15, 17), albeit with a significant minority showing a completely ablated CD8 response to Omicron peptide pools (17). Other studies show that around a fifth of responders to peptide panels have a 50-70% drop in T cell response (16). Our approach was to evaluate T cell recognition using the dual approach of mapped epitope pools spanning the mutated regions and also, whole, naturally processed antigen. We found the greatest impairment of T cell recognition when looking at epitope recognition after processing of whole antigen. Naturally processed epitopes from uptake of whole antigen would generally be considered more representative of the real-life situation and nearer to HLA-ligandome studies than synthetic megapools of several hundred overlapping peptides which have the potential to drown out physiological response patterns under the noise of responses from cryptic epitopes that may not feature in real-life natural responses. That is, megapool approaches can, by their nature, underestimate the extent of response ablation. The natural HLA-ligandome of peptides shown to be elicited by natural processing and HLAII presentation only partially overlaps epitopes mapped from overlapping synthetic peptide panels (34, 35). Our immunization of mice with B.1.1.529 mutant epitopes confirmed that de novo T cell response repertoire can be elicited, but this is not necessarily the same as that generated during live infection.
In summary, these studies have shown that the high global prevalence of B.1.1.529 (Omicron) infections and reinfections likely reflects considerable subversion of immune recognition at both the B, T cell, antibody binding and nAb level, although with considerable differential modulation through immune imprinting. Some imprinted combinations, such as infection during the Wuhan Hu-1 and Omicron waves, confer particularly impaired responses.


Dr. M

Reynolds Science